Progress in Development Studies 20, 2 (2020) pp.

119–139


Wealth inequality and inter-governorate
migration: Evidence from Egypt
Mohamed Arouri
Université Nice Sophia Antipolis, France

Nguyen Viet Cuong

Institute of Theoretical and Applied Research (ITAR), Duy Tan
University, Da Nang, Vietnam

Thang Long Institute of Mathematics and Applied Sciences, Thang

Long University, Hanoi, Vietnam

Abstract: Migration is one of the key livelihood strategies for households, especially those in low-
income and middle-income regions. In this study, we investigate whether the economic level and
inequality of wealth can affect inter-governorate migration in Egypt. Using gravity models and data
from Population and Housing Censuses of Egypt, we are able to measure the push as well as pull
effects of economic and wealth inequality levels on internal migration flows. Although there are
a large number of studies on the effect of economic levels on migration, there is little if anything
known about the effect of wealth inequality between the origin and destination areas on migration.
We measure wealth levels using the household asset index. We find that people tend to move to
governorates with high wealth levels as well as high wealth inequality. There is a positive association
between wealth inequality and economic growth in Egypt. Governorates with high wealth inequal-
ity tend to experience high economic growth and therefore attract more migrants. This study’s
findings also suggest that unlike non-work migration, the low wealth level in origin governorates is
a push factor for work migration.

Key words: Household asset inequality, household asset index, migration, gravity model, Egypt

I. Introduction efficiency in terms of poverty reduction largely

Although economic growth constitutes a depends on the level of wealth inequality.
necessary condition to reduce poverty, there Nowadays, most countries pursue a pro-
is an agreement among economists that its poor growth policy that not only promotes

© 2020 Sage Publications 10.1177/1464993420910554

120 Wealth inequality and inter-governorate migration: Evidence from Egypt
economic growth but also reduces wealth
inequality (Bhagwati, 1988; Goudie and Ladd,
1999; Kakwani and Ernesto, 2000; Perkins
et al., 2001).
Migration is an important livelihood
strategy for households, especially in low-
income and middle-income regions, to increase
income and reduce poverty. Economists
as well as policy makers have long been
interested in understanding the causes of
migration. In the economic literature, there are
numerous theories on the migration decisions
of individuals or households. In conventional
theory, individuals relocate to maximize utility,
given spatial variation in wage and price levels
(Molloy et al., 2011; Valencia, 2008). According
to the New Economics of Labor Migration
(NELM), decisions to migrate depend on
the characteristics of both migrants and their
families (Stark and Bloom, 1985; Stark and
Taylor, 1991). There are several recent studies
showing that community characteristics
of home and destination locations are also
important factors exerting ‘push’ and ‘pull’
forces on migrants (Ackah and Medvedev
2012; Mayda, 2007; Kim and Cohen, 2010).
People tend to move from regions with low
welfare levels to those with high welfare levels
(Stark, 1991; Stark and Bloom, 1985; Stark
and Taylor, 1991). In addition, public services
are also considered in migrants’ decisions.
Migrants seek not only better employment
but also access to improved public services in
destination areas (Zaiceva and Zimmermann,
2008). Ackah and Medvedev (2012) show
that in communities with poor public services,
people are likely to migrate regardless of their
relatively disadvantaged education and/or
inherent characteristics that are not favorable
for migration.
Human capital plays a key role in economic
development. Understanding whether welfare
levels can attract migrants, especially highly
educated and skilled migrants, is important for
economic development. The main objective
of this study is to examine the push and pull
effects of economic levels and wealth inequality
Progress in Development Studies 20, 2 (2020) pp. 119–139
levels on inter-governorate migration using a
gravity model and data from the Population
and Housing Censuses of Egypt in 1996 and
2006. We provide a descriptive analysis of
inter-governorate migration in Egypt and then
examine whether the mean and inequality of
the origin and destination governorates can
affect inter-governorate migration.
T here are no data on incomes or
expenditures in the Population and Housing
Censuses of Egypt. To measure the wealth
level, we use the household asset index
approach. Especially, we follow Filmer and
Pritchett (2001) to compute a household
asset index, which is an aggregate measure of
assets of households, including durable goods,
housing characteristics, and access to utilities.1
T his ar ticle aims to make several
contributions to the related literature. First, it
provides the first empirical findings on the push
and pull effects of economic levels on migration
in Egypt. Although there is a consensus that
individuals tend to migrate from less developed
areas to more developed ones (Stark and
Bloom, 1985; Stark and Taylor, 1991), we
address the remaining need to measure
the magnitude of the effect of economic
development on migration (Taylor, 2006).
Second, although the effect of migration on
wealth inequality of the origin and destination
areas has been studied (e.g., Black et al.,
2005; Card and Shleifer, 2009), there is little
if anything known about the effect of wealth
inequality between the origin and destination
areas on migration. The effect of wealth
inequality between the source and destination
areas on migration flows is captured by the
push and pull effects of the wealth levels of
the respective areas. This study is one of the
first attempts to examine the impact of wealth
inequality between the origin and destination
areas on migration flows. Policy makers are
interested in assessing the impact of region-
wise inequality not only for economic reasons
but also for social ones such as social justice
and social cohesion.

For several reasons, Egypt offers an
interesting case to look at. First, Egypt is the
largest country by population in the Arab
world. It is a lower middle-income country
with a per capita Gross Domestic Product
(GDP) of around US$3,300 as in 2013 (World
Bank, 2014). Second, Egypt has achieved an
annual economic growth rate of around 5
percent, but it has not been very successful
in poverty reduction. Poverty in Egypt is
persistent, with the rate around 20 percent
during the last two decades (El-Laithy, 2011).
According to the World Bank (2014), the
poverty rate of Egypt is 25.2 percent, and 75
percent of the poor are living in rural areas.
There is high wealth inequality between as well
as within regions. It is important to investigate
whether wealth inequality and economic levels
are pull or push factors for migration among
governorates in Egypt. Third, although there
are several studies on migration in Egypt such
as Zohry (2009), Herrera and Badr (2012),
Wahba (2015), and Arouri and Nguyen (2018),
there are no studies on the linkage between
migration and wealth inequality in Egypt.
The remaining sections of this article
are structured as follows. Section II briefly
discusses the theoretical framework and
literature on the determinants of migration
decisions. Section III introduces the main
data sets used in this study and presents some
preliminary analyses. Section IV presents our
empirical approach. Section V discusses our
main empirical findings. Finally, conclusions
are presented in Section VI.
II. Theoretical framework and literature
review
According to theoretical migration models
(e.g., Harris and Todaro, 1970), migration is
determined by ‘pull’ and ‘push’ factors. The
most important pull factors are economic
incentives such as job opportunities, higher real
wages and better public services. Push factors
at the place of origin such as poor economic
activity or conflicts cause out-migration. This
‘disequilibrium’ view of migration emphasizes
Progress in Development Studies 20, 2 (2020) pp. 119–139
Arouri and Cuong 121
persistent expected income differentials as
a major motivation for migration. Recent
research tries to identify factors behind
migration, taking into account market failures
due to information asymmetries, credit market
imperfections, and network effects.
In the empirical literature, logistic models
are often used to understand migrants’ motives
for moving (Gibson et al., 2011; McKenzie
et al., 2010). These models include characteri-
stics of migrants as well as of their households
and home and host regions. Microeconomic
as well as gravity models are used in the
literature to identify migration determinants.
Microeconomic models make use of micro data
such as surveys of individuals or households.
On the other hand, gravity models appeal to
the representative agent assumption and make
use of aggregate data, for example, census
data in which migration rates are measured
at the level of the community or the admini-
strative unit (e.g., Grogger and Hanson, 2011;
Karemera et al., 2000; Ortega and Peri, 2013;
Phan and Coxhead, 2010). For example, Phan
and Coxhead (2010) used data from the 1989
and 1999 censuses from Vietnam to investi-
gate migration patterns and determinants
and the role of migration on cross-province
income differentials. They found that provinces
with a higher per capita income attract more
migrants. However, the coefficient of the
income in the sending province is also positive
and significant, implying that the ‘liquidity
constraint effect’ outweighed the ‘push’ effect
in inhibiting migration from poorer regions.
It is well established in the literature that
individuals move from low to high wealth
areas (Stark and Bloom, 1985; Stark and
Taylor, 1991). However, there is less evidence
on the effect of inequality of wealth between
the origin and destination areas on migration.
Lipton (1980) suggests that wealth inequalities
within origin areas can push migration, that is,
people living in areas that are more unequal are
more likely to migrate.
Based on the current literature on econo-
mic growth, poverty, inequality, and migration,
122 Wealth inequality and inter-governorate migration: Evidence from Egypt
we argue that wealth inequality might affect
migration between areas through several
channels. First, income inequality, which has
been shown to be linked to migration, can be
caused by asset inequality. McKinley (1993)
suggests that unequal and highly concentrated
distribution of assets, in particular productive
resources such as physical capital, land, and
human capital in the form of better education,
cause unequal distribution of income.
Second, capital market imperfections
and credit constraints coupled with unequal
and highly concentrated patterns of asset
ownership can constitute a push factor of
migration. In the case of Egypt, it seems
that the rural poor are not able to provide
the collateral to obtain sufficient loans from
lenders. In the context of credit rationing, only
entrepreneurs with significant levels of assets
are able to finance their projects. This prevents
the rural poor from undertaking profitable
investments, may lead to the persistence of
poverty, and may encourage the rural poor
to migrate to urban areas to find alternative
means of livelihood (Deininger and Olinto,
2000).
Third, asset inequality limits human capital
accumulation and thus growth, especially in
countries such as Egypt where agriculture
activity is still economically important due to
its contribution to the GDP, rural employment
and livelihoods for the poor (Carter, 2000).
This may explain the high rate of rural-to-
urban migration among less educated people.
Finally, asset inequality may threaten social
cohesion and cause social instability and
social stratification, which can be directly
associated with violence and crime and may
act as a migration push factor (De Janvry and
Sadoulet, 2000).
III. Data sets and descriptive statistics
Data sets
In this article, we make use of data from the
Population and Housing Censuses of Egypt
that were conducted by the Central Agency
Progress in Development Studies 20, 2 (2020) pp. 119–139
for Mobilization and Statistics (CAPMAS) in
1996 and 2006. We use the 10 percent sample
of the censuses. The numbers of individuals in
the 1996 and 2006 data sets are 5,902,243 and
7,282,434, respectively.
The Population and Housing Censuses of
Egypt contain individual-level and household-
level data. The individual-level module
includes data on demography, education,
employment, disability, and migration of
individuals. Regarding migration, there are
data on the previous governorate of individuals
living in the current governorate and their year
of moving. Using the data set, we estimate
the rate of inter-governorate migration. The
household-level module contains data on
housing condition facilities and durables. These
data are used to construct the household asset
indexes.
Inter-governorate migration
This article uses the Egypt Population and
Housing Censuses to compute the inter-
governorate migration and estimate the
migration flows between governorates.
Migration tended to decrease over the
1996–2006 period. Figure 1 shows that the
proportion of individuals who changed their
place of residence in the past 10 years was
3.2 percent in 1996 and 2.1 percent in 2006.
The proportion of people moving across
governorates was slightly lower, at 2.1 percent
in 1996 and 2.0 percent in 2006. In this study,
we measure the effect of the wealth level and
the inequality of wealth on migration flows.
We measure inter-governorate migration
during the past 2 years. The percentage of
migrants during the past 2 years was lower
than the percentage of migrants during the
past 10 years, since more people moved in
the past 10 years than in the past 2 years. The
percentages of inter-governorate migration
during the past 2 years in 1996 and 2006 were
0.39 percent and 0.38 percent, respectively.
Figures 2 and 3 present maps of the
percentage of the out-migration and
in-migration population during the past
 Arouri and Cuong 123

Figure 1. Migration rate in Egypt

Source: Authors’ estimation from the 1996 and 2006 Egypt Population and Housing Censuses.

Figure 2. The percentage of inter-governorate in-migration and out-migration during

the past 2 years in 1996
Source: Authors’ estimation from the 1996 Egypt Population and Housing Census.

Progress in Development Studies 20, 2 (2020) pp. 119–139

124 Wealth inequality and inter-governorate migration: Evidence from Egypt

Figure 3. The percentage of inter-governorate in-migration and out-migration during

the past 2 years in 2006
Source: Authors’ estimation from the 2006 Egypt Population and Housing Census.

2 years across governorates. The pattern of approach, an index is constructed as the first
migration is very similar in 1996 and 2006. principal component of a vector of assets of
South Sinai had the highest in-migration households, including durable goods, housing
rate, at 5.6 percent, while Kaliobia had the characteristics, and access to utilities. Filmer
lowest in-migration rate, at 0.16 percent. The and Scott (2008) and Kolenikov and Angeles
governorates that had the highest in-migration (2009) conclude that the rankings of various
rates were Red Sea and New Valley.
An important aspect is the motive for
migration. Table 1 presents the distribution of Table 1. Distribution of migrants by the
migrants by the reasons for their migration. reasons for migration and migrants’ age
It shows that two-thirds of migration across
(in %)
governorates in Egypt is due to family
movements and marriages. Work migrants Reason for migration 1996 2006
accounted for 18.2 percent and 21.0 percent Work 18.22 21.01
of the total migrants in 1996 and 2006, Family move 31.68 31.81
respectively. Since migration mainly happens Marriage or union 36.22 36.17
for family reasons, migrants are mainly children Divorce or widowhood 0.52 0.59
and young people (aged 15–30). Older people Study 1.43 1.95
are less likely to move. In addition, migration Other reasons 11.92 8.46
is also a human capital investment, and older Total 100 100
workers have a shorter period to collect By age
migration investment returns and so have a Children (aged 0–14) 21.80 27.17
lower incentive to migrate (Borjas, 2012). Young people (aged 15–30) 51.90 49.95
Adults (aged 31–60) 24.15 20.99
Household Asset Index Elderly (aged 61+) 2.15 1.89
We follow the principal components approach Total 100 100
of Filmer and Pritchett (2001) to compute Source: Authors’ estimation from the 1996 and 2006
a household asset index. According to this Egypt Population and Housing Censuses.

Progress in Development Studies 20, 2 (2020) pp. 119–139


measures of welfare, including outcomes for
education, health care, fertility, child mortality,
and the labor market, are very similar to
the ranking of asset indices. The principal
component approach defines a household asset
index in terms of the first principal component
of the variables used. The household asset
index for household j, denoted by A j, is
computed as follows:
x pj - xr p
A j = | ap e o(1)
p Sp
where x p denotes asset p, and xr denotes
the mean of households in the sample. s is
the standard deviation of asset xp, and the
p-dimensional vector of weight a is chosen to
maximize the sample variance of A, provided
| a2p = 1. Weight a is also called the vector
p
of scores of asset variables, which can be
estimated using principal component analysis.
The average of the household asset index is
equal to zero. The index has both negative and
positive values. Inequality measures such as the
Gini and Generalize Entropy indices are widely
used for calculating income and consumption.
However, these measures are used for positive
values and cannot be applied in the case of the
household asset index. Instead, in this study,
we use the asset inequality measure proposed
by McKenzie (2005) to measure the household
asset inequality of governorates in Egypt.
The household asset inequality of a local
area such as a governorate can be measured
by calculating the standard deviation of the
household asset index across households in
the local area. McKenzie (2005) proposes a
relative asset inequality measure of an area,
which is the ratio of the standard deviation of
the household asset index in the area to the
standard deviation in the sample as a whole.
More specifically, the relative measure of
household asset inequality of a governorate
is given by:
SD governate ( A )
Ij = ,(2)
SD all ( A )
Progress in Development Studies 20, 2 (2020) pp. 119–139
Arouri and Cuong 125
where SDgovernate(A) is the standard deviation
of the household asset index, A, in the
governorate, and SD all(A) is the standard
deviation of household asset index A in the
whole country. This asset inequality measure
is positive and can be larger than one. A
higher value of the asset inequality measure
of an area means higher inequality in assets
among the households within the area.
McKenzie (2005) proves this asset inequality
measure satisfies the common properties of
an inequality measure. It should be noted that
since we standardize the household asset index
according to the standard normal distribution,
the inequality index is equal to the standard
deviation of the household asset index.
Using data from the 1996 and 2006 Egypt
Population and Housing Censuses, we can
compute the household asset index. Based
on the availability of data, we use dummy
variables indicating whether individuals live in a
household with the following assets or housing
conditions: piped water, septic tank latrine,
house, house with at least one bathroom,
house with at least one kitchen, electric cooker,
computer, car, washing machine, fridge or
refrigerator, television, and video player. Figure
A1 presents the proportion of individuals
living in households with these assets. Using
these variables, we can compute the average
household asset index and the inequality of
the household asset index of the governorates
(Table A1).
Figure 4 depicts the geographic maps of
the average household asset index by quintiles.
Governorates are grouped in quintiles from
the lowest to highest level of household asset
index. The green governorates have the
highest value of household asset index, while
the brown and red governorates have the
lowest value of household asset index. Figure
4 shows that some governorates like Red Sea
and South Sinai improved their average wealth
during the period 1996–2006. Figure 5 presents
the inequality pattern. The green color depicts
low inequality levels, while brown color depicts
high inequality levels.
126 Wealth inequality and inter-governorate migration: Evidence from Egypt

Figure 4. Average household asset index

Source: Authors’ estimation from the 1996 and 2006 Egypt Population and Housing Censuses.

Figure 5. Inequality of household asset index

Source: Authors’ estimation from the 1996 and 2006 Egypt Population and Housing Censuses.

IV. Estimation Method and Rotte, 2000). Gravity models assume

In this study, we aim to measure both
push and pull effects of wealth level and
inequality on migration. In the literature of
migration, gravity models of migration are
often used to estimate the flow of migration
between geographic areas (e.g., Grogger
and Hanson, 2011; Hatton and Williamson,
2002; Karemera et al., 2000; Ortega and
Peri, 2013; Phan and Coxhead, 2010; Volger
that migration flows between two areas are
directly proportional to their size (population
or GDP) and are inversely proportional to
the physical distance between them (see,
e.g., Poot et al., 2016; Ramos Lobo, 2016).
Thus, we also make use of a gravity model
to estimate the push and pull impacts of the
economic level and inequality of origin and
destination governorates on inter-governorate

Progress in Development Studies 20, 2 (2020) pp. 119–139


migration in Egypt. First, we write the basic
gravity model as follows:
P ia, t P jb, t
M i, j, t = g f p ,(3)
D cij
where M i,j,t is the migration flow from
governorate i to governorate j in year t (i and
j can be any two geographic areas, but in our
study they are governorates); Pi,t and Pj,t are
the sizes of the populations of governorates
i and j in year t, respectively; and Dij is the
distance between the two governorates.
Taking the log of both sides of Equation (1),
we get:
log ( M ijt) = log ( g ) + a log ( Pi, t) +
(4)
b log ( Pj, t) - c log ( D i, j).
To model the effect of asset level and
inequality, we include variables of level and
inequality of household assets in governorates
i and j as follows:
log ( M ij, t) = b 0 + b 1 A i, t + b 2 A j, t +
b 3 I i, t + b 4 I j, t + b 5 log ( Pi, t) +
b 6 log ( Pj, t) + b 7 log ( D i, j) +
b 8 T t + f i, j, t (5)
where Ai,t and Ai,t are the variables of the
average level of household asset index of
governorates i and j in year t, respectively.
Ii,t and Ij,t are the variables of the inequality of
household asset index of governorates i and j in
year t, respectively. Tt is a year dummy, which
is equal to one for 2006 and zero for 1996. fi,j,t
are conventional random error terms.
The push and pull effects of the (average)
level of household assets on migration
are measured by the coefficients b 1 and
b 2, respectively, while the push and pull
effects of the level of household asset
inequality are measured by the coefficients
b3 and b4, respectively. If people move from
governorates with low levels of household
assets to governorates with high levels of
household assets, b1 < 0 and b2 > 0. Similarly,
if people move from governorates with high
inequality of household assets to governorates
Progress in Development Studies 20, 2 (2020) pp. 119–139
Arouri and Cuong 127
with low inequality of household assets, b3 >
0 and b4 < 0.
In non-experimental analysis, selection
bias or endogeneity bias is always a great
challenger. Although we cannot fully address
this bias, we try to mitigate it in several ways.
First, we include governorate dummies in
regression, which can eliminate endogeneity
bias caused by time-invariant unobserved
variables. Second, we use the lagged variables
of assets to measure current migration to
avoid reverse causality. More specifically, the
migration between governorates is measured
using the 2006 Population and Housing
Census of Egypt, and the level and inequality
of household assets are measured using the
1996 Population and Housing Census of
Egypt. Third, although we do not have data
on the assets of households before migration,
we exclude migrants when computing the
household asset index of governorates to
avoid reverse causality. In addition, we only
measure the effect of asset level and inequality
on migration during the past 2 years.
A potential problem in the estimation
model (5) is that the dependent variable can
be zero because of no migration between
governorate pairs. As a result, we cannot take
the logarithm. In this study, we use two-part
models (Duan et al., 1983; Manning et al.,
1987). In the first part, we model the dummy
variables of migration happening between
governorates, and in the second part, we only
apply model (5) to the governorates that have
migration flows between them.
For sensitivity analysis, we use an
additional estimator. We transform Equation
(5) into an exponential function and apply the
Poisson pseudo-maximum likelihood estimator
(Gourieroux et al., 1984; Santos Silva and
Tenreyro, 2006). The results from the Poisson
pseudo-maximum likelihood model, which
are reported in Table A2, are very similar
to the two-part model. In this article, we
use the results from the two-part model for
interpretation.
128 Wealth inequality and inter-governorate migration: Evidence from Egypt
V. Empirical Findings
In this section, we discuss the push and pull
effects of asset and wealth inequality levels
on inter-governorate migration in Egypt.
The dependent variables are the log of the
number of migrants across governorates. In
addition to overall migration, we also measure
the effect on different types of migration,
including migration of people who have higher
education, migration for different purposes
and migration of different age groups. As
mentioned in the previous section, we focus
on the inter-governorate migration flow during
the last 2 years.
Table 2 presents the two-part regressions
on the migration flows. There is no migration
between nearly 40 percent of paired
governorates. Geographic distance increases
the cost of migration, and people are more
likely to move to a closer governorate. If the
distance between two governorates increases
by 1 percent, the probability of migration
between the two governorates decreases by
0.12 percent. For governorate pairs with inter-
governorate migration, a 1 percent increase in
the distance between governorates reduces
the number of migrants by 0.45 percent. As
expected, migration flows are higher between
governorates with higher populations.
Regarding asset and wealth inequality
levels, people tend to move to the governorates
with high assets and high asset inequality.
Areas with high economic levels and high
wealth inequality attract more migrants than
areas with low economic levels and low wealth
inequality. If the asset and wealth inequality
levels of destination governorates increase
by 0.1 standard deviation, the probability of
migration to these governorates increases by
0.022 and 0.056 percent, respectively. Their
effect on the migration flows is high. A 0.1
standard deviation increase in the asset level
and wealth inequality index of destination
governorates is associated with 10.5 percent
and 25.3 percent increases in the number of
immigrants, respectively.
It is well established that people tend to
move to areas with high income (Stark and
Progress in Development Studies 20, 2 (2020) pp. 119–139
Bloom, 1985; Stark and Taylor, 1991; Stark,
1991). Thus, this explains why governorates
with high wealth levels attract more migrants.
However, it is more difficult to explain why
people tend to move to governorates with
high wealth inequality levels. According
to Kuznets (1955), there is an inverted-U
relationship between the level of a country’s
economic growth and its wealth inequality
levels. Empirical findings on this association
are inconclusive (e.g., see review from De
Groot et al., 2008; Turnovsky, 2013). To
test this relation for the case of Egypt, we
graph the growth rate of the wealth level
of governorates over the 1996–2006 period
across their degree of wealth inequality in
1996. Figure A2 shows a negative correlation
between the wealth growth rate and initial
inequality level. The correlation coefficient is
negative and statistically significant at the 10
percent level. It suggests that areas with high
initial inequality levels tend to experience a high
rate of wealth growth, which in turn attracts
more immigrants. This might be a possible
explanation for why high inequality is a pull
factor of migration in Egypt.
There are no data on poverty status from
the censuses. However, there is information
on the education level of people. Table 2 also
reports the effect of asset levels and wealth
inequality on the migration of people with
higher education (i.e., those who completed
college or university). The effect of distance on
the migration of people with higher education
is smaller than the effect of distance on overall
migration. This implies that distance is less
challenging for migrants with higher education
than for general migrants. Interestingly, people
with higher education tend to move more
from, as well as to, governorates with high
asset levels and wealth inequality than other
people. In other words, high asset levels and
wealth inequality are not only pull but also push
factors for migrants with higher education.
Table 3 presents the effects of asset level
and wealth inequality on migration for different
purposes. There are some differences in
Table 2. Regression of migration flows
Having higher education Log of higher education
Explanatory variables Having migration Log of migration flow migration migration flow
Log of distance –0.1225*** –0.4499*** –0.1218*** –0.1776***
(0.0113) (0.0419) (0.0125) (0.0451)
Log of population of origin governorates 0.1631*** 0.5230*** 0.1285*** 0.4498***
(0.0069) (0.0289) (0.0068) (0.0353)
Log of population of destination governorates 0.0526*** 0.3630*** 0.0524*** 0.2870***
(0.0068) (0.0274) (0.0071) (0.0290)
Level of household asset index of origin 0.0204 –0.0549 0.0913*** 0.3067***
governorates
(0.0160) (0.0641) (0.0181) (0.0687)
Level of household asset index of destination 0.2239*** 1.0560*** 0.3101*** 0.7364***
governorates
(0.0160) (0.0655) (0.0171) (0.0702)
Inequality of household asset index of origin 0.0352 0.1720 0.1999*** 0.4154*
governorates
(0.0435) (0.2162) (0.0433) (0.2459)
Inequality of household asset index of 0.5636*** 2.5303*** 0.7774*** 2.1536***
destination governorates
(0.0438) (0.1781) (0.0478) (0.1846)
Year 2006 –0.1525*** –0.3090*** –0.1177*** 0.0564
(0.0176) (0.0692) (0.0173) (0.0723)
Constant –2.1920*** –8.3135*** –2.3229*** –8.6434***
(0.1978) (0.8421) (0.2121) (0.9534)
Observations 2,808 1,733 2,808 956
R-squared 0.308 0.336 0.278 0.327
Source: Authors’ estimation from the 1996 and 2006 Egypt Population and Housing Censuses.
Notes: Robust standard errors in parentheses.
***p < 0.01, **p < 0.05, *p < 0.1.
Table 3. Regression of flows of migration by purposes
Log of other-
Having work Log of work Having family Log of family Having marriage Log of marriage Having other- type migration
Explanatory variables migration migration flow migration migration flow migration migration flow type migration flow
Log of distance –0.0980*** –0.1557*** –0.1201*** –0.2679*** –0.1457*** –0.3458*** –0.1340*** –0.2980***
(0.0128) (0.0462) (0.0126) (0.0489) (0.0123) (0.0418) (0.0125) (0.0437)
Log of population of 0.1532*** 0.5034*** 0.1422*** 0.4389*** 0.1221*** 0.3854*** 0.1179*** 0.2838***
origin governorates (0.0071) (0.0325) (0.0070) (0.0366) (0.0070) (0.0328) (0.0071) (0.0316)
Log of population 0.0459*** 0.2444*** 0.0536*** 0.3003*** 0.0925*** 0.3639*** 0.1027*** 0.4138***
of destination (0.0072) (0.0279) (0.0072) (0.0292) (0.0069) (0.0281) (0.0068) (0.0286)
governorates
Level of household –0.0230 –0.2015*** 0.0676*** –0.1606** 0.0857*** 0.0190 0.0557*** –0.0380
asset index of origin (0.0185) (0.0679) (0.0184) (0.0746) (0.0184) (0.0687) (0.0182) (0.0698)
governorates
Level of household 0.3180*** 0.7512*** 0.3192*** 0.8003*** 0.2663*** 0.7737*** 0.2461*** 0.4866***
asset index (0.0172) (0.0722) (0.0172) (0.0725) (0.0172) (0.0662) (0.0174) (0.0667)
of destination
governorates
Inequality of 0.0375 -0.1826 0.1263*** 0.2194 0.1131** 0.1223 0.0896* 0.3258
household asset index (0.0447) (0.2482) (0.0451) (0.2486) (0.0442) (0.2446) (0.0459) (0.2175)
of origin governorates
Inequality of 0.6721*** 2.0541*** 0.7192*** 1.6743*** 0.5836*** 1.6194*** 0.6059*** 1.5703***
household asset (0.0483) (0.1796) (0.0478) (0.1865) (0.0475) (0.1792) (0.0465) (0.1757)
index of destination
governorates
Year 2006 –0.1373*** –0.2212*** –0.1219*** –0.1694** –0.1235*** –0.3032*** –0.0608*** –0.2889***
(0.0180) (0.0727) (0.0180) (0.0799) (0.0174) (0.0709) (0.0180) (0.0733)
Constant –2.3853*** –8.0769*** –2.3455*** –6.9915*** –2.4064*** –7.2032*** –2.5699*** –6.6326***
(0.2157) (0.8971) (0.2122) (0.9716) (0.2171) (0.8954) (0.2168) (0.8835)
Observations 2,808 1,205 2,808 1,181 2,808 992 2,808 1,051
R-squared 0.290 0.249 0.285 0.256 0.297 0.336 0.271 0.264
Source: Authors’ estimation from the 1996 and 2006 Egypt Population and Housing Censuses.
Notes: Robust standard errors in parentheses.
***p < 0.01, **p < 0.05, *p < 0.1.
Table 4. Regression of flows of migration by age
Having Log of middle-
Having child Log of child Having young Log of young middle-age age migration Having elderly Log of elderly
Explanatory variables migration migration flow migration migration flow migration flow migration migration flow
Log of distance –0.1077*** –0.2582*** –0.1221*** –0.4224*** –0.1295*** –0.1912*** –0.0592*** –0.0996*
(0.0128) (0.0464) (0.0120) (0.0398) (0.0125) (0.0447) (0.0104) (0.0514)
Log of population of origin 0.1423*** 0.3860*** 0.1579*** 0.4681*** 0.1380*** 0.4426*** 0.0613*** 0.2778***
governorates (0.0069) (0.0353) (0.0072) (0.0276) (0.0070) (0.0325) (0.0055) (0.0423)
Log of population of 0.0542*** 0.2740*** 0.0671*** 0.3427*** 0.0550*** 0.3138*** 0.0656*** 0.1558***
destination governorates (0.0073) (0.0276) (0.0071) (0.0264) (0.0073) (0.0270) (0.0061) (0.0291)
Level of household 0.0663*** –0.1178* 0.0211 –0.1950*** 0.0415** 0.1022 0.0424*** 0.1609**
asset index of origin (0.0184) (0.0708) (0.0175) (0.0611) (0.0184) (0.0664) (0.0152) (0.0799)
governorates
Level of household asset 0.2810*** 0.7766*** 0.2353*** 0.8326*** 0.3152*** 0.7544*** 0.2098*** 0.1676**
index of destination (0.0174) (0.0697) (0.0168) (0.0630) (0.0172) (0.0680) (0.0155) (0.0744)
governorates
Inequality of household 0.1448*** 0.2024 0.0784* –0.1348 0.0991** 0.3162 0.0658* 0.4217
asset index of origin (0.0452) (0.2382) (0.0460) (0.2042) (0.0450) (0.2310) (0.0354) (0.3080)
governorates
Inequality of household 0.6990*** 1.8078*** 0.5910*** 2.2736*** 0.5870*** 1.7336*** 0.3936*** 0.4049*
asset index of destination (0.0483) (0.1732) (0.0465) (0.1722) (0.0499) (0.1727) (0.0392) (0.2101)
governorates
Year 2006 –0.0949*** –0.1908** –0.1715*** –0.2453*** –0.1060*** –0.2914*** –0.0409*** –0.2549***
(0.0179) (0.0753) (0.0182) (0.0666) (0.0179) (0.0703) (0.0134) (0.0840)
Constant –2.4836*** –6.2499*** –2.4469*** –7.3201*** –2.1438*** –8.1347*** –1.7052*** –3.6800***
(0.2147) (0.9319) (0.2103) (0.8126) (0.2161) (0.9009) (0.1921) (1.0815)
Observations 2,808 1,058 2,808 1,541 2,808 1,122 2,808 375
R-squared 0.265 0.264 0.290 0.331 0.285 0.285 0.191 0.181
Source: Authors’ estimation from the 1996 and 2006 Egypt Population and Housing Censuses.
Notes: Robust standard errors in parentheses.
***p < 0.01, **p < 0.05, *p < 0.1.
Children are aged less than 15, the young people aged 15–30; middle age people aged 31–60, and elderly aged above 60.
132 Wealth inequality and inter-governorate migration: Evidence from Egypt
incentives to migrate for different migration
purposes. First, people tend to move to a richer
and higher-inequality governorate regardless
of the purpose of migration. However, for
marriage and family migration, high inequality
within the origin area is also a push factor for
migration. For work migration, the low level of
assets is a push factor. In other words, working
people tend to move from a governorate of a
low economic level to another governorate of
a high economic level. The economic incentive
is more important for work migrants than for
other migrants. In Table 4, we investigate the
effects of wealth level and inequality on the
inter-governorate migration flows of different
age groups. The effect is very similar between
groups, with people tending to move to a
governorate with high asset levels and high
wealth inequality.
VI. Conclusions
In this study, we look at the effect of the
average household asset index and the
inequality in household asset indices on inter-
governorate migration in Egypt using gravity
models and data from the 1996 and 2006
Population and Housing Censuses of Egypt.
To do this, we follow Filmer and Pritchett
(2001) and construct an household asset index
as the first principal component of a vector of
assets of households, including durable goods,
housing characteristics, and access to utilities.
Our main findings show that people are
more likely to move to a closer governorate.
Progress in Development Studies 20, 2 (2020) pp. 119–139
As expected, migration flows are higher
between governorates with higher populations.
Regarding asset levels and wealth inequality,
people tend to move to the governorates with
high average asset levels as well as high asset
inequality levels. Areas with high economic
levels and high asset inequality attract more
migrants than areas with low economic levels
and low asset inequality. For work migration,
a low average level of household assets is a
push factor. In other words, working people
tend to move from a governorate of a low
economic level to another governorate of a
high economic level.
All in all, our empirical findings suggest
that economic inequality acts as a pull factor
for internal migration. Thus, Egyptian policy
makers might continue to aim not only at
rural poverty reduction but also at inequality
reduction. As assets become more equally
distributed and increase in level, internal
migration and rural poverty will tend to
decrease. Future research works should
explore the best approaches in the case of
Egypt towards reducing asset inequality and,
ultimately, reducing rural poverty.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with
respect to the research, authorship, and/or publication of
this article.
Funding
The authors received no financial support for the research,
authorship, and/or publication of this article.
 Arouri and Cuong 133

Appendix A

Figure A1. Proportion of individuals living in households with different assets

Source: Authors’ estimation from the 1996 and 2006 Egypt Population and Housing Censuses.

Progress in Development Studies 20, 2 (2020) pp. 119–139

134 Wealth inequality and inter-governorate migration: Evidence from Egypt

Figure A2. Wealth growth and inequality

Source: Authors’ estimation from the 1996 and 2006 Egypt Population and Housing Censuses.

Table A1. Mean and inequality of household asset index

1996 2006
Inequality of Inequality of
Mean of household household asset Mean of household household asset
Governorates asset index index asset index index
Cairo 0.843 0.686 0.752 1.163
Alexandria 0.829 0.659 0.492 0.927
Port Said 0.966 0.712 0.503 1.126
Suez 0.965 0.520 0.665 0.863
Damietta 0.561 0.606 0.315 0.653
Dakahlia 0.261 0.694 0.261 0.677
Sharkia 0.023 0.783 0.014 0.774
Kaliobia 0.173 0.825 0.140 0.798
Kafr Sheikh –0.252 0.805 –0.032 0.632
Gharbia 0.078 0.808 0.125 0.744
Menoufia –0.325 0.801 –0.208 0.827
Behera –0.234 0.793 –0.128 0.655
(Table A1 continued)

Progress in Development Studies 20, 2 (2020) pp. 119–139

 Arouri and Cuong 135

(Table A1 continued)
1996 2006
Inequality of Inequality of
Mean of household household asset Mean of household household asset
Governorates asset index index asset index index
Ismailia 0.412 0.800 0.259 0.840
Giza 0.405 0.901 0.281 1.008
Bani Swif –0.804 0.952 –0.754 1.042
Fayoum –0.988 1.038 –0.443 0.896
Menia –1.016 0.932 –0.783 0.966
Asiut –0.631 0.927 –0.530 0.930
Sohag –0.671 0.874 –0.427 0.855
Qena –0.452 0.864 –0.549 0.867
Aswan 0.101 0.798 –0.133 0.822
Luxor 0.074 0.764 –0.005 0.851
Red Sea 0.236 0.892 –0.520 1.602
New Valley 0.464 0.680 0.031 0.902
Marsa Matroh –0.350 0.832 –0.483 0.947
North Sinai –0.252 1.074 –0.497 1.208
South Sinai 0.065 1.195 –1.260 1.732
All the country 0 1 0 1
Source: Authors’ estimation from the 1996 and 2006 Egypt Population and Housing Censuses.

Progress in Development Studies 20, 2 (2020) pp. 119–139

Table A2. Poisson pseudo maximum likelihood regression of migration
Total Higher Work Family Marriage Other-type Child Young Middle-age Elderly
Explanatory migration education migration migration migration migration migration migration migration migration
variables flow flow flow flow flow flow flow flow flow flow
Log of distance –0.5993*** –0.4015* –0.4091*** –0.6707*** –0.7110*** –0.5982*** –0.6483*** –0.6275*** –0.4927*** –0.4677***
(0.1607) (0.2120) (0.1051) (0.1847) (0.2378) (0.1424) (0.1655) (0.1506) (0.1832) (0.1678)
Log of 1.2061*** 1.3401*** 1.2101*** 1.2419*** 1.3705*** 0.9941*** 1.2396*** 1.1430*** 1.3236*** 1.2727***
population (0.1060) (0.1199) (0.0951) (0.1252) (0.1640) (0.0949) (0.1159) (0.1002) (0.1294) (0.1303)
of origin
governorates
Log of 0.6412*** 0.6897*** 0.4727*** 0.6156*** 0.8911*** 0.8255*** 0.6190*** 0.6336*** 0.6802*** 0.7274***
population of (0.0666) (0.1090) (0.0517) (0.0759) (0.0969) (0.0682) (0.0715) (0.0618) (0.0783) (0.0736)
destination
governorates
Level of 0.2315 0.9859*** –0.2650* 0.2600 0.7583*** 0.2738 0.1659 0.0837 0.5921*** 0.6933***
household asset (0.1943) (0.2172) (0.1517) (0.2168) (0.2559) (0.1982) (0.2024) (0.1934) (0.2066) (0.1853)
index of origin
governorates
Level of 1.1768*** 1.2972*** 1.2237*** 1.2747*** 1.3534*** 0.7529*** 1.2497*** 1.0968*** 1.2937*** 1.1678***
household (0.0837) (0.1033) (0.0995) (0.0966) (0.0997) (0.1105) (0.0931) (0.0812) (0.0982) (0.1338)
asset index
of destination
governorates
Inequality of 1.5918* 1.9215** 0.4229 1.8315* 2.0484* 1.8691** 1.9751** 1.4976* 1.3179 0.8704
household asset (0.8614) (0.9089) (0.6468) (0.9401) (1.0594) (0.7668) (0.8415) (0.8560) (0.9440) (0.8693)
index of origin
governorates
Inequality of 3.0696*** 4.0005*** 3.2977*** 2.8181*** 3.1637*** 3.1716*** 3.1686*** 3.1248*** 2.8491*** 2.2681***
household (0.3569) (0.3594) (0.3060) (0.4134) (0.5524) (0.3133) (0.3738) (0.3373) (0.4174) (0.4149)
asset index
of destination
governorates
(Table A2 continued)
(Table A2 continued)

Total Higher Work Family Marriage Other-type Child Young Middle-age Elderly
Explanatory migration education migration migration migration migration migration migration migration migration
variables flow flow flow flow flow flow flow flow flow flow
Year 2006 –0.7719*** –0.6724*** –0.6370*** –0.6195*** –1.2918*** –0.8260*** –0.6018*** –0.7896*** –0.8846*** –0.7023***
(0.1919) (0.1942) (0.1440) (0.2331) (0.2814) (0.1703) (0.2136) (0.1804) (0.2113) (0.1979)
Constant –22.625*** –29.425*** –21.877*** –23.416*** –30.396*** –24.158*** –24.491*** –22.026*** –26.608*** –28.271***
(2.8429) (4.1877) (2.2514) (3.2206) (4.2658) (2.7615) (3.0302) (2.6083) (3.4618) (3.0888)
Observations 2,808 2,808 2,808 2,808 2,808 2,808 2,808 2,808 2,808 2,808
R-squared 0.194 0.232 0.211 0.170 0.210 0.209 0.195 0.191 0.205 0.222
Source: Authors’ estimation from the 1996 and 2006 Egypt Population and Housing Censuses.
Notes: Robust standard errors in parentheses.
***p < 0.01, **p < 0.05, *p < 0.1.
Children are aged less than 15, the young people aged 15–30; middle age people aged 31–60, and elderly aged above 60.
138 Wealth inequality and inter-governorate migration: Evidence from Egypt
Note
1. Filmer and Scott (2008) and Kolenikov and Angeles
(2009) conclude that the rankings of various measures
of welfare, including outcomes for education, health
care, fertility, child mortality, and the labor market,
are very similar to the ranking of asset indices.
References
Ackah, C. and Medvedev, D. 2012: Internal migration
in Ghana: Determinants and welfare impacts.
International Journal of Social Economics 39(10),
764–784.
Arouri, M. and Nguyen, C.V. 2018: Does international
migration affect labor supply, non-farm diversification
and welfare of households? Evidence from Egypt.
International Migration 56(1), 39–62.
Bhagwati, J.N. 1988: Poverty and public policy. World
Development 16(5), 539–654.
Black, R., Natali, C. and Skinner, J. 2005: Migration
and inequality. World Bank.
Borjas, G.J. 2012: Labor economics (6th ed.). McGraw
Hill/Irwin.
Card, D. and Shleifer, A. 2009: Immigration and
inequality. American Economic Review 99(2), 1–21.
Carter, M. 2000: Fertility of Mexican immigrant women
in the US: A closer look. Social Science Quarterly
81(1), 404–420.
De Groot, H.L.F., de Dominicis, L. and Florax,
R.J.G.M. 2008: Growth and inequality: A meta-
analysis. Scottish Journal of Political Economy 55(5),
654–682.
de Janvry, A. and Sadoulet, E. 2000: Growth, poverty,
and inequality in Latin America: A causal analysis,
1970–94. The Review of Income and Wealth 46(3),
267–288.
Deininger, K. and Olinto, P. 2000: Asset distribution,
inequality and growth (World Bank Research WP
2375). Washington, D.C.: World Bank. Available
at: http://documents.worldbank.org/curated/
en/408611468766839670/pdf/multi-page.pdf
Duan, N., Manning, W.G., Moris, C., and
Newhouse, J. P. 1983: A comparison of alternative
models for the demand for medical care. Journal of
Business and Economics Statistics 1(2), 115–126.
El-Laithy, H. 2011: The ADCR 2011: Poverty in
Egypt, United Nations development programme,
Arab development challenges report (Background
Paper 2011/11). New York, NY: United Nations
Development Programme. Available at: https://
www.undp.org/content/dam/rbas/doc/poverty/
BG_11_Poverty%20in%20Egypt.pdf
Filmer, D. and Pritchett, L.H. 2001: Estimating
wealth effects without expenditure data-or tears:
An application to educational enrollments in states
of India. Demography 38(1), 115–132.
Progress in Development Studies 20, 2 (2020) pp. 119–139
Filmer, D. and Scott, K. 2008: Assessing asset indices.
The World Bank.
Gibson, J., McKenzie, D. and Stillman, S. 2011:
The impacts of international migration on remaining
household members: Omnibus results from a
migration lottery program. Review of Economics and
Statistics 93(4), 1297–1318.
Gourieroux, C., Monfort, A., and Trognon, A. 1984:
Pseudo maximum likelihood methods: Applications to
Poisson models. Econometrica 52(3), 701–720.
Goudie, A. and Ladd, P. 1999: Economic growth,
poverty and inequality. Journal of International
Development 11, 177–195.
Grogger, J. and Hanson, G.H. 2011: Income
maximization and the selection and sorting of
international migrants. Journal of Development
Economics 95(1), 42–57.
Harris, J.R. and Todaro, M.P. 1970: Migration,
unemployment and development: A two-sector
analysis. American Economic Review 60(1), 126–142.
Hatton, T. and Williamson, J.G. 2002: What
fundamentals drive world migration? (NBER Working
Paper No. 9159). National Bureau of Economic
Research.
Herrera, S. and Badr, K. 2012: Internal migration in
Egypt levels, determinants, wages, and likelihood of
employment (Policy Research Working Paper No.
6166). The World Bank.
Kakwani, N. and Ernesto, P.M. 2000: What is pro-
poor growth. Asian Development Review 18(1), 1–16.
Karemera, D., Oguledo, V.I. and Davis, B. 2000: A
gravity model analysis of international migration to
North America. Applied Economics 32(13), 1745– 1755.
Kim, K. and Cohen, J. 2010: Determinants of interna-
tional migration flows to and from industrialized
countries: A panel data approach beyond gravity.
International Migration Review 44(4), 899–932.
Kolenikov, S. and Angeles, G. 2009: Socioeconomic
status measurement with discrete proxy variables: Is
principal component analysis a reliable answer? The
Review of Income and Wealth 55, 128–165.
Kuznets, S. 1955: Economic growth and income
inequality. American Economic Review 45, 1–28.
Lipton, M. 1980: Migration from rural areas of poor
countries: The impact on rural productivity and
income distribution. World Development 8, 1–24.
Manning, W. G., Duan, N. and Rogers, W.H. 1987:
Monte Carlo evidence on the choice between
sample selection and two-part models. Journal of
Econometrics 35(1), 59–82.
Mayda, A. 2007: International migration: A panel data
analysis of the determinants of bilateral flows (CReAM
Discussion Paper No 07/07). Centre for Research
and Analysis of Migration Department of Economics,
University College London.
McKenzie, D. 2005: Measuring inequality with asset
indicators. Journal of Population Economics 18(2),
229–260.

McKenzie, D., Stillman, S. and Gibson, J. 2010:
How important is selection? Experimental vs.
non-experimental measures of the income gains
from migration. Journal of the European Economic
Association 8(4), 913–945.
McKinley, T. 1993: The distribution of wealth in
rural China. In: Griffin K and Renwei Z (eds) The
Distribution of Income in China. London: Macmillan
Press, pp. 116–134.
Molloy, R., Smith, C. and Wozniak, A. 2011: Internal
migration in The United States (NBER Working
Paper 17307). National Bureau of Economic
Research.
Ortega, F. and Peri, G. 2013: The effect of income
and immigration policies on international migration.
Migration Studies 1(1), 47–74.
Perkins, D.H., et al. 2001: Economics of development.
W.W. Norton & Company, Inc.
Phan, D. and Coxhead, I. 2010: Interprovincial migration
and inequality during Vietnam’s transition. Journal of
Development Economics 91(1), 100–112.
Poot, J., et al. 2016: The gravity model of migration: The
successful comeback of an ageing superstar in regional
science (IZA Discussion Paper No. 10329). The
Institute for the Study of Labor (IZA).
Ramos Lobo, R. 2016: Gravity models: A tool for
migration analysis. IZA World of Labor 239, 1–10.
Santos Silva, J. M. C. and Tenreyro, S. 2006: The
log of gravity. The Review of Economics and Statistics
88(4), 641–658.
Stark, O. 1991. The migration of labour. Cambridge:
Harvard University Press.
Progress in Development Studies 20, 2 (2020) pp. 119–139
Arouri and Cuong 139
Stark, O. and Bloom, D. 1985: The new economics
of labor migration. American Economic Review 75,
173–178.
Stark, O. and Taylor, J. 1991: Migration incentives,
migration types: The role of relative deprivation. The
Economic Journal 101, 1163–1178.
Taylor, E. 2006: International migration and economic
development. Department of Economic and Social
Affairs, United Nations.
Turnovsky, S.J. 2013: The relationship between
economic growth and inequality. New Zealand
Economic Papers 47(2), 113–139.
Valencia, J.M. 2008: Migration and its determinants:
A study of two communities in Colombia. Atlantic
Economic Journal 36(2), 247–260.
Volger, M. and Rotte, R. 2000: The effects of
development on migration: Theoretical issues and
new empirical evidence. Journal of Population
Economics 13(3), 485–508.
Wahba, J. 2015: Selection, selection, selection: The
impact of return migration. Journal of Population
Economics 28(3), 535–563.
World Bank. 2014: World development indicators
database 2014. The World Bank.
Zaiceva, A. and Zimmermann, K. 2008: Scale,
diversity, and determinants of labour migration in
Europe (IZA Discussion Paper No. 3595). Born,
Germany: The Institute for the Study of Labor.
Available at: http://ftp.iza.org/dp3595.pdf
Zohry, A. 2009: The development impact of internal
migration: Findings from Egypt. International Union
for the Scientific Study of Population, XXVI IUSSP
International Population Conference.